oadmin/sub2api
1
0
Fork 0
Code Issues Pull Requests 1 Actions 3 Packages Projects Releases Wiki Activity

feat: 区分 Anthropic 5m/1h 缓存创建 token 的差异化计费

Browse Source 
Anthropic API 的 cache_creation 对象区分了 ephemeral_5m 和 ephemeral_1h
两种缓存创建 token，1h 单价远高于 5m（如 claude-3-5-haiku: 5m=$1/MTok,
1h=$6/MTok）。此前系统统一按 5m 单价计费，导致计费偏低。

后端：
- pricing_service: 加载 LiteLLM 的 cache_creation_input_token_cost_above_1hr
- billing_service: GetModelPricing 启用分类计费（安全守卫 1h>5m），
  CalculateCost 按 5m/1h 分别计费，无明细时回退到 5m 单价
- gateway_service: parseSSEUsage/handleNonStreamingResponse 用 gjson
  提取嵌套 cache_creation 对象的 ephemeral_5m/1h_input_tokens
- antigravity_gateway_service: extractSSEUsage/extractClaudeUsage 同步提取
- usage_log: 修复 GORM column tag 确保写入正确的数据库列
- 新增迁移 054: 删除 GORM 自动生成的重复列

前端：
- 使用记录 tooltip 展示 5m/1h 缓存创建明细（带彩色 badge 区分）
- 表格单元格缓存写入数值旁显示 1h 标识
This commit is contained in:
shaw
2026-02-14 18:15:35 +08:00
parent 2857fa2ef7
commit a817cafe3d
10 changed files with 174 additions and 42 deletions
Download Patch File Download Diff File Expand all files Collapse all files

50
backend/internal/service/gateway_service.go
View File

@@ -349,6 +349,8 @@ type ClaudeUsage struct {
OutputTokens int `json:"output_tokens"`
CacheCreationInputTokens int `json:"cache_creation_input_tokens"`
CacheReadInputTokens int `json:"cache_read_input_tokens"`
CacheCreation5mTokens int // 5分钟缓存创建token（来自嵌套 cache_creation 对象）
CacheCreation1hTokens int // 1小时缓存创建token（来自嵌套 cache_creation 对象）
}
// ForwardResult 转发结果
@@ -4401,6 +4403,14 @@ func (s *GatewayService) parseSSEUsage(data string, usage *ClaudeUsage) {
usage.InputTokens = msgStart.Message.Usage.InputTokens
usage.CacheCreationInputTokens = msgStart.Message.Usage.CacheCreationInputTokens
usage.CacheReadInputTokens = msgStart.Message.Usage.CacheReadInputTokens
// 解析嵌套的 cache_creation 对象中的 5m/1h 明细
cc5m := gjson.Get(data, "message.usage.cache_creation.ephemeral_5m_input_tokens")
cc1h := gjson.Get(data, "message.usage.cache_creation.ephemeral_1h_input_tokens")
if cc5m.Exists() || cc1h.Exists() {
usage.CacheCreation5mTokens = int(cc5m.Int())
usage.CacheCreation1hTokens = int(cc1h.Int())
}
}
// 解析message_delta获取tokens（兼容GLM等把所有usage放在delta中的API）
@@ -4429,6 +4439,14 @@ func (s *GatewayService) parseSSEUsage(data string, usage *ClaudeUsage) {
if msgDelta.Usage.CacheReadInputTokens > 0 {
usage.CacheReadInputTokens = msgDelta.Usage.CacheReadInputTokens
}
// 解析嵌套的 cache_creation 对象中的 5m/1h 明细
cc5m := gjson.Get(data, "usage.cache_creation.ephemeral_5m_input_tokens")
cc1h := gjson.Get(data, "usage.cache_creation.ephemeral_1h_input_tokens")
if cc5m.Exists() || cc1h.Exists() {
usage.CacheCreation5mTokens = int(cc5m.Int())
usage.CacheCreation1hTokens = int(cc1h.Int())
}
}
}
@@ -4449,6 +4467,14 @@ func (s *GatewayService) handleNonStreamingResponse(ctx context.Context, resp *h
return nil, fmt.Errorf("parse response: %w", err)
}
// 解析嵌套的 cache_creation 对象中的 5m/1h 明细
cc5m := gjson.GetBytes(body, "usage.cache_creation.ephemeral_5m_input_tokens")
cc1h := gjson.GetBytes(body, "usage.cache_creation.ephemeral_1h_input_tokens")
if cc5m.Exists() || cc1h.Exists() {
response.Usage.CacheCreation5mTokens = int(cc5m.Int())
response.Usage.CacheCreation1hTokens = int(cc1h.Int())
}
// 兼容 Kimi cached_tokens → cache_read_input_tokens
if response.Usage.CacheReadInputTokens == 0 {
cachedTokens := gjson.GetBytes(body, "usage.cached_tokens").Int()
@@ -4566,10 +4592,12 @@ func (s *GatewayService) RecordUsage(ctx context.Context, input *RecordUsageInpu
} else {
// Token 计费
tokens := UsageTokens{
InputTokens: result.Usage.InputTokens,
OutputTokens: result.Usage.OutputTokens,
CacheCreationTokens: result.Usage.CacheCreationInputTokens,
CacheReadTokens: result.Usage.CacheReadInputTokens,
InputTokens: result.Usage.InputTokens,
OutputTokens: result.Usage.OutputTokens,
CacheCreationTokens: result.Usage.CacheCreationInputTokens,
CacheReadTokens: result.Usage.CacheReadInputTokens,
CacheCreation5mTokens: result.Usage.CacheCreation5mTokens,
CacheCreation1hTokens: result.Usage.CacheCreation1hTokens,
}
var err error
cost, err = s.billingService.CalculateCost(result.Model, tokens, multiplier)
@@ -4603,6 +4631,8 @@ func (s *GatewayService) RecordUsage(ctx context.Context, input *RecordUsageInpu
OutputTokens: result.Usage.OutputTokens,
CacheCreationTokens: result.Usage.CacheCreationInputTokens,
CacheReadTokens: result.Usage.CacheReadInputTokens,
CacheCreation5mTokens: result.Usage.CacheCreation5mTokens,
CacheCreation1hTokens: result.Usage.CacheCreation1hTokens,
InputCost: cost.InputCost,
OutputCost: cost.OutputCost,
CacheCreationCost: cost.CacheCreationCost,
@@ -4747,10 +4777,12 @@ func (s *GatewayService) RecordUsageWithLongContext(ctx context.Context, input *
} else {
// Token 计费（使用长上下文计费方法）
tokens := UsageTokens{
InputTokens: result.Usage.InputTokens,
OutputTokens: result.Usage.OutputTokens,
CacheCreationTokens: result.Usage.CacheCreationInputTokens,
CacheReadTokens: result.Usage.CacheReadInputTokens,
InputTokens: result.Usage.InputTokens,
OutputTokens: result.Usage.OutputTokens,
CacheCreationTokens: result.Usage.CacheCreationInputTokens,
CacheReadTokens: result.Usage.CacheReadInputTokens,
CacheCreation5mTokens: result.Usage.CacheCreation5mTokens,
CacheCreation1hTokens: result.Usage.CacheCreation1hTokens,
}
var err error
cost, err = s.billingService.CalculateCostWithLongContext(result.Model, tokens, multiplier, input.LongContextThreshold, input.LongContextMultiplier)
@@ -4784,6 +4816,8 @@ func (s *GatewayService) RecordUsageWithLongContext(ctx context.Context, input *
OutputTokens: result.Usage.OutputTokens,
CacheCreationTokens: result.Usage.CacheCreationInputTokens,
CacheReadTokens: result.Usage.CacheReadInputTokens,
CacheCreation5mTokens: result.Usage.CacheCreation5mTokens,
CacheCreation1hTokens: result.Usage.CacheCreation1hTokens,
InputCost: cost.InputCost,
OutputCost: cost.OutputCost,
CacheCreationCost: cost.CacheCreationCost,